Total and beverage-specific alcohol use at mid-life and all-cause and cause-specific mortality: a prospective cohort study

Abstract

Background

While detrimental health impacts of excessive alcohol use are well-established, potential effects of low-to-moderate drinking remain controversial. We examined the associations between total and beverage-specific alcohol intake and mortality, including cardiovascular disease (CVD), cancer (e.g., 20 site-specific cancers), and other mortality.

Methods

We included 340,924 participants from UK Biobank (2006–2022). Alcohol use was categorized as: never or occasional alcohol consumption (≤ 20 g/week), low consumption (> 20 g/week to ≤ 20 g/d for men, > 20 g/week to ≤ 10 g/d for women), moderate consumption (> 20 to ≤ 40 g/d for men, > 10 to ≤ 20 g/d for women), and high consumption (> 40 g/d for men and > 20 g/d for women). We used Cox proportional hazard models to examine the associations of alcohol intake and mortality.

Results

During a mean follow-up of 13.4 years, we documented 22,381 deaths. Compared with never or occasional alcohol intake, high total alcohol intake was associated with higher mortality from all-causes (HR and 95%CI: 1.24 (1.17, 1.31)), CVD (1.14 (1.01, 1.28)), cancer (1.36 (1.26, 1.47)), and other causes (1.12 (1.02, 1.22)), whereas moderate alcohol use was associated with higher cancer mortality (1.11 (1.03, 1.20)) only. Alcohol consumption from spirits and beer/cider, even at low to moderate levels, was associated with higher mortality from all-causes, CVD, cancer, and other diseases (HRs ranging from 1.07 to 1.83). Yet, low and moderate alcohol intake from wine was associated with lower all-cause, CVD, cancer, or other mortality. Importantly, moderate to high intake of total alcohol and alcohol from beer/cider and spirits, and high wine use were particularly associated with higher mortality from multiple site-specific cancers, including mortality from head/neck, respiratory system, gastrointestinal, liver/gallbladder, hematological, female breast, and female/male genital cancers.

Conclusions

The results suggested that low to moderate alcohol intake from spirits and beer/cider was associated with higher all-cause, CVD, cancer, and other mortality, whereas low to moderate alcohol use from wine was associated with lower all-cause, CVD, cancer, and other mortality. However, high alcohol use, regardless of beverage-type, mortality from various site-specific cancers was elevated.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12937-026-01361-3.

Introduction

Heavy alcohol consumption is one of the leading causes of morbidity and mortality, representing a major public health problem [1]. However, epidemiological evidence on the health impact of low to moderate alcohol consumption is mixed, and whether low to moderate alcohol consumption is neutral or even beneficial to health continues to be a contentious topic [2–5]. Importantly, recently, the dietary guidelines, alcohol guidelines and public health agencies in multiple countries, including US, UK, and Canada have consistently recommended low- to moderate- dose consumption as generally safe [6–8]. However, more recently, WHO has stated no safe amount for health [9]. In early 2025, the US Surgeon General’s Advisory recommended zero alcohol consumption for preventing cancers [10]. In June 2025, a Scientific Statement from the American Heart Association has highlighted that it remains unclear whether alcohol consumption is part of a healthy lifestyle [11]. These contrasting messages have caused much public confusion regarding whether low to moderate alcohol consumption is harmful or not.

To clarify the potential health impact of low to moderate alcohol intake, important knowledge gaps need to be addressed. First, whether and how low to moderate alcohol intake is associated with all-cause and cause-specific premature deaths, particularly cardiovascular (CVD) and cancer deaths remains uncertain [2–5]. Second, the health impact of alcohol may vary by types of alcoholic beverages. For example, the antioxidants in wine may have a favorable impact on health, whereas the purines in beer could exert an adverse impact [12]. Yet, evidence linking types of alcohol to mortality is limited and mixed [13–18]. Finally, most prior studies have focused on only total cancer mortality, with mixed results [4, 19], despite substantial variance in the underlying etiologies and mechanisms across various specific cancers. Yet, the association between alcohol intake and site-specific cancer mortality remains unclear. Overall, addressing these gaps is essential for informing evidence-based dietary recommendations.

To address these gaps, we examined the associations between total and beverage-specific alcohol intake (e.g., red, white, and fortified wine, beer/cider, and spirits) with all-cause and cause-specific mortality (e.g., CVD, cancer, and other mortality) among adults ≥ 40 years from the UK Biobank (UKB), particularly including 20 site-specific cancer mortality. In addition, we investigated the associations between timing of alcohol with respect to meals with mortality, given that alcohol timing affects acetaldehyde absorption and bioavailability, potentially influencing organ-specific toxicity [20].

Methods

Study design and participants

The UKB is a prospective cohort of 502,507 participants aged 40–70 years across the UK, with initial enrolment in 2006–2010 (baseline) and follow-up from 2012 [21]. The National Information Governance Board for Health and Social Care and National Health Service Northwest Multicenter Research Ethics Committee approved the protocols of the UKB. Written informed consent was obtained from all participants. After excluding participants without complete alcohol data at baseline (n = 109,452), former drinkers (n = 255), baseline pregnant women (n = 90), and those with CVD/cancer at baseline (n = 50,488), or without follow-up data (n = 1,298), 340,924 participants remained for the analyses (Supplementary Fig. 1).

Alcohol intake assessment

Alcohol intake was collected using an alcohol-frequency questionnaire at baseline. Participants reported consumption frequency and average amounts of alcohol types. We computed alcohol intake (g/week) using beverage-specific quantities multiplied by their standard ethanol equivalents (e.g., 1 unit-equivalent described as containing 8 g of pure alcohol; 125 mL wine = 1.6 unit-equivalents, 1 pint of beer (586 mL) = 2.6 unit-equivalents, 25 mL spirits = 1 unit-equivalent, 62.5 mL fortified wine = 1 unit-equivalent) [20]. Per the recommendation from health authorities and previous studies [4, 22–24], we categorized alcohol consumption into 4 categories: never or occasional consumption: ≤20 g/week for men and women (reference); low consumption: >20 g/week to ≤ 20 g/d for men, > 20 g/week to ≤ 10 g/d for women; moderate consumption: >20 g/d to ≤ 40 g/d for men, > 10 g/d to ≤ 20 g/d for women; high consumption: >40 g/d for men and > 20 g/d for women. Given relatively narrow range for spirits consumption, we categorized alcohol from spirits into 3 categories: never or occasional, low, and moderate to high consumption [20]. Additional classifications included: alcohol consumption with meals (yes/no/varied), and beverage preference (wine preference (> 80% of alcohol from wine), other, or none) [4].

Ascertainment of deaths

Deaths were identified from the death certificates via the National Health Service (NHS) Information Center (England/Wales) and the NHS Central Register (Scotland) through December 2022. The 10th revision of the International Classification of Diseases (ICD-10) was used to identify the underlying cause of death (I00-I99 and C00-C97 for CVD and cancer deaths, respectively). Supplementary Table 1 presents information to confirm site-specific cancer mortality (e.g., head/neck cancers).

Assessment of covariates

Information on sociodemographic and lifestyle, including age, sex, race, education, smoking, Townsend deprivation index (TDI), and physical activity was collected using questionnaires at baseline. TDI ranged from − 6.26 to 10.16, with higher score indicating greater socioeconomic deprivation [25]. The frailty index (FI) was computed as the proportion of 49 health deficits present (range: 0–1), with higher scores indicating greater frailty [26, 27]. Details on these covariates are provided in Supplementary Methods.

Statistical analysis

We used Cox proportional hazards models to estimate the hazard ratios (HRs) and 95%CIs for associations of total and beverage-specific alcohol intake (e.g., wine, beer/cider, and spirits), with all-cause, CVD, cancer, and other mortality. The proportional hazards assumption was tested using Schoenfeld residuals [28], and no violation was detected. Covariates were selected a priori based on prior literature, biological plausibility, and their potential role as confounders in the association between alcohol intake and mortality. Specifically, we adjusted for age, sex, race, the regions of assessment area centers, smoking, education, TDI, healthy eating index-2020 (HEI-2020), physical activity, blood glucose, pressure, and cholesterol, BMI, FI, family history of diabetes, CVD, and cancer, alcoholic beverage preference, and alcohol consumption with meals in analyses. We also examined the associations of alcohol intake from more specific wine (e.g., red, white, and fortified wine) with all-cause and cause-specific mortality. Alcohol intake from red, white, and fortified wine, beer/cider, and spirits were mutually adjusted. We assessed linear trends by modeling median alcohol intake per category as a continuous variable. We further examined the associations of total and beverage-specific alcohol intake with 20 site-specific cancer mortality (e.g., mortality from head/neck cancers, respiratory system cancers, gastrointestinal cancers, liver/gallbladder cancers, thyroid cancer, female breast cancer, bone/soft tissue cancers, kidney/bladder cancers, hematological malignancies, and female/male genital cancers). Given multiple tests being performed for site-specific cancer mortality, P values were corrected for false discovery rate (FDR) using the Benjamini–Hochberg method; an FDR-corrected P < 0.05 was considered statistically significant.

We fitted restricted cubic spline analyses with four knots to assess the dose-response relationships between total and beverage-specific alcohol intake with all-cause and cause-specific mortality [29]. We examined the associations of timing of alcohol respective to meals with all-cause and cause-specific mortality and the joint associations of amount and timing of alcohol intake with mortality.

Subgroup and sensitivity analyses

We conducted stratified analyses for alcohol intake and mortality by age, sex, race, smoking, physical activity, HEI-2020, or BMI. We examined the interactions by adding interaction terms of alcohol intake and the risk factors in fully adjusted model. In sensitivity analyses, we excluded participants who died within the first two years of the follow-up, as well as lifetime abstainers and non-drinkers, to reduce potential bias from pre-existing illness and reverse causation. In addition, we conducted competing risk analyses using cause-specific hazard models for CVD mortality, cancer mortality, and other-cause mortality, treating deaths from other causes as competing events.

We used the random-forest method to impute the missing data of these covariates [30]. We conducted statistical analyses using R version 4.4.1, with P < 0.05 considered statistically significant.

Results

Baseline characteristics of participants

Compared with participants with never or occasional intake of total alcohol, those with high intake of total alcohol were more physically active (Table 1). Compared with participants with never or occasional intake of alcohol from beer/cider, spirits, and wine, respectively, those with high intake of alcohol from beer/cider were more likely to be men and more physically active, and had lower diet quality and educational attainment; those with moderate and high intakes of alcohol from spirits were more physically active and had lower diet quality and educational attainment; and those with high intake of alcohol from wine were more likely to be women and white, and had higher diet quality (Supplementary Table 2).

Table 1.

Baseline characteristics of participants per total alcohol consumption

	Total alcohol consumption
	Overall (n = 340,924)	Never or occasional (n = 35,958)	Low (n = 120,528)		Moderate (n = 107,976)	High (n = 76,462)
Total alcohol consumption (g/week)	148.6 (144.8)	10.2 (4.9)	64.7 (32.4)	148.1 (60.0)		330.0 (183.3)
Age (years)	55.9 (8.0)	56.1 (8.2)	56.2 (8.1)		55.8 (8.0)	55.4 (7.9)
Gender (%)
- Female	51.2	74.6	44.5		51.6	52.3
Race (%)
- White	96.3	88.5	95.8		97.6	98.2
Physical activity (MET-min/week)	2,689.5(2,706.9)	2,674.2 (2,726.1)	2,639.4 (2,659.9)		2,662.3 (2,639.2)	2,812.0 (2,859.9)
Smoking status (%)
- Current smoker	10.4	10.4	10.7		10.5	10.2
- Former smoker	34.3	34.4	34.0		34.3	34.3
- Never smoker	55.3	55.2	55.3		55.2	55.5
HEI-2020 (0-100)	48.9 (9.0)	48.1 (9.3)	48.8 (8.9)		49.1 (8.9)	48.9 (9.2)
BMI (kg/m2)	27.1 (4.4)	27.6 (5.3)	26.9 (4.4)		26.8 (4.2)	27.2 (4.4)
Frailty index (0–1)	0.08 (0.05)	0.10 (0.06)	0.08 (0.05)		0.08 (0.05)	0.09 (0.05)
Education (%)
- College or above	11.6	11.7	11.3		11.8	11.9
- High school	36.6	32.6	37.8		37.7	34.3
- Middle school and below	51.8	55.7	50.9		50.5	53.8
Drinking with meals (%)
- No	20.6	19.6	19.0		19.3	24.5
- Yes	43.8	54.5	49.3		43.0	42.9
- Varied	35.6	25.9	31.7		37.7	32.6

Data are mean (SD) and percentages

Never or occasional consumption: ≤20 g/week for men and women; low consumption: >20 g/week to ≤ 20 g/d for men, and > 20 g/week to ≤ 10 g/d for women; moderate consumption: > 20 to ≤ 40 g/d for men, and > 10 to ≤ 20 g/d for women; high consumption: > 40 g/d for men, and > 20 g/d for women

The frailty index was calculated using a cumulative deficit model by summing the presence of 49 health-related deficits and dividing by the total number of deficits considered, which ranged from 0 to 1, with higher values indicating more deficits and greater frailty, and lower values reflecting better overall health

MET Metabolic equivalent of task, HEI Healthy Eating Index, BMI Body mass index

Associations of total and beverage-specific alcohol intake with mortality

During a mean follow-up of 13.4 years, 22,381 deaths (4,288 CVD deaths, 11,063 cancer deaths, and 7,030 other deaths) occurred among 340,924 participants. After multi-variable adjustments, compared with never or occasional intake of total alcohol, high intake of total alcohol was associated with higher all-cause mortality (HR and 95%CI: 1.24 (1.17, 1.31)), CVD mortality (1.14 (1.01, 1.28)), cancer mortality (1.36 (1.26, 1.47)), and other mortality (1.12 (1.02, 1.22)); moderate intake of total alcohol was associated with higher cancer mortality (1.11 (1.03, 1.20)), but not with all-cause, CVD, and other mortality (Table 2).

Table 2.

Associations of total alcohol intake with all-cause and cause-specific mortality

Total alcohol consumption	Never or occasional (n = 35,958)	Low (n = 120,528)	Moderate (n = 107,976)	High (n = 76,462)	P for trend
All-cause mortality
No of deaths	2,051	7,531	6,826	5,973
Model 1	1.00 (reference)	0.86 (0.81, 0.90)	0.92 (0.88, 0.97)	1.21 (1.15, 1.28)	< 0.0001
Model 2	1.00 (reference)	0.94 (0.91, 1.01)	1.02 (0.97, 1.07)	1.24 (1.17, 1.31)	< 0.0001
CVD mortality
No of deaths	388	1,519	1,224	1,157
Model 1	1.00 (reference)	0.83 (0.74, 0.93)	0.81 (0.72, 0.92)	0.81 (0.72, 0.92)	< 0.0001
Model 2	1.00 (reference)	0.94 (0.83, 1.06)	0.91 (0.80, 1.02)	1.14 (1.01, 1.28)	< 0.0001
Cancer mortality
No of deaths	965	3,615	3,488	2,995
Model 1	1.00 (reference)	0.92 (0.86, 0.99)	1.04 (0.97, 1.12)	1.34 (1.24, 1.44)	< 0.0001
Model 2	1.00 (reference)	0.99 (0.91, 1.06)	1.11 (1.03, 1.20)	1.36 (1.26, 1.47)	< 0.0001
Other mortality
No of deaths	698	2,397	2,114	1,821
Model 1	1.00 (reference)	0.77 (0.71, 0.84)	0.82 (0.75, 0.89)	1.06 (0.97, 1.16)	< 0.0001
Model 2	1.00 (reference)	0.88 (0.80, 0.97)	0.94 (0.86, 1.03)	1.12 (1.02, 1.22)	< 0.0001

Never or occasional consumption: ≤20 g/week for men and women; low consumption: >20 g/week to ≤ 20 g/d for men, and > 20 g/week to ≤ 10 g/d for women; moderate consumption: > 20 to ≤ 40 g/d for men, and > 10 to ≤ 20 g/d for women; high consumption: > 40 g/d for men, and > 20 g/d for women

Model 1: Age (continuous, years), sex (male, female), race (white, non-white), and the regions of assessment center (Manchester, Liverpool, Newcastle, Middlesbrough, Leeds, Sheffield, Birmingham, Nottingham, Cambridge, Norwich, London, Oxford, Reading, Bristol, Plymouth, Glasgow, Edinburgh, Cardiff, Swansea)

Model 2: Model 1 + smoking status (never, former, current), education (middle school and below, high school, college or above), Townsend Deprivation Index (continuous), HEI-2020 (continuous), physical activity (continuous, MET-min/week), blood pressure (continuous, mmHg), blood glucose (continuous, mmol/l), blood cholesterol (continuous, mmol/l), family history of diabetes (yes, no), family history of CVD (yes, no), family history of cancer (yes, no), BMI (continuous, kg/m²), frailty index (continuous), alcoholic beverage preference (wine preference, other preference, no preference), alcohol drinking with meals (yes, no, varied)

The positive associations with mortality were mainly driven by alcohol intake from beer/cider and spirits with a linear trend (all P_{linear trend} <0.05) (Table 3). Compared with never or occasional intake of alcohol from beer/cider, high alcohol intake from beer/cider was associated with higher all-cause mortality (1.51 (1.43, 1.61)), CVD mortality (1.47 (1.29, 1.67)), cancer mortality (1.66 (1.52, 1.80)), and other mortality (1.34 (1.21, 1.49)). Moderate intake of alcohol from beer/cider was still associated with all-cause mortality (1.17 (1.11, 1.25)), cancer mortality (1.25 (1.16, 1.35)), and other mortality (1.15 (1.05, 1.26)). Even low intake of alcohol from beer/cider was associated with higher cancer mortality (1.07 (1.01, 1.13)) (Table 3). Similarly, compared with never or occasional intake of alcohol from spirits, moderate and high intake of alcohol from spirits was associated with higher all-cause mortality (1.62 (1.53, 1.73)), CVD mortality (1.51 (1.30, 1.74)), cancer mortality (1.55 (1.41, 1.69)), and other mortality (1.83 (1.64, 2.03)). Even low intake of alcohol from spirits was associated with higher all-cause mortality (1.08 (1.04, 1.12)), CVD mortality (1.09 (1.01, 1.18)), cancer mortality (1.07 (1.02, 1.12)), and other mortality (1.09 (1.02, 1.16)) (Table 3). In contrast, compared with never or occasional intake of alcohol from wine, low intake of alcohol from wine was associated with lower mortality from all-cause (0.89 (0.85, 0.92)), CVD (0.86 (0.78, 0.94)), cancer (0.92 (0.87, 0.97)), and other causes (0.86 (0.81, 0.93)); moderate intake of alcohol from wine was associated with lower mortality from all-cause (0.92 (0.88, 0.97)), CVD (0.79 (0.70, 0.88)), and other causes (0.87 (0.80, 0.95)); while high intake of alcohol from wine was associated with higher cancer mortality (1.10 (1.02, 1.20)) (Table 3).

Table 3.

Associations of alcohol consumption from beer/cider, spirits and wine with all-cause and cause-specific mortality

Alcohol consumption from beer/cider	Never or occasional	Low	Moderate	High	P for trend
All-cause mortality	1.00 (reference)	1.01 (0.97, 1.04)	1.17 (1.11, 1.25)	1.51 (1.43, 1.61)	< 0.0001
CVD mortality	1.00 (reference)	1.00 (0.92, 1.09)	1.02 (0.90, 1.15)	1.47 (1.29, 1.67)	< 0.0001
Cancer mortality	1.00 (reference)	1.07 (1.01, 1.13)	1.25 (1.16, 1.35)	1.66 (1.52, 1.80)	< 0.0001
Other mortality	1.00 (reference)	0.92 (0.86, 1.01)	1.15 (1.05, 1.26)	1.34 (1.21, 1.49)	< 0.0001

Alcohol consumption from spirits	Never or occasional	Low	Moderate and high		P for trend
All-cause mortality	1.00 (reference)	1.08 (1.04, 1.12)	1.62 (1.53, 1.73)		< 0.0001
CVD mortality	1.00 (reference)	1.09 (1.01, 1.18)	1.51 (1.30, 1.74)		< 0.0001
Cancer mortality	1.00 (reference)	1.07 (1.02, 1.12)	1.55 (1.41, 1.69)		< 0.0001
Other mortality	1.00 (reference)	1.09 (1.02, 1.16)	1.83 (1.64, 2.03)		< 0.0001

Alcohol consumption from wine	Never or occasional	Low	Moderate	High	P for trend
All-cause mortality	1.00 (reference)	0.89 (0.85, 0.92)	0.92 (0.88, 0.97)	1.02 (0.96, 1.07)	0.28
CVD mortality	1.00 (reference)	0.86 (0.78, 0.94)	0.79 (0.70, 0.88)	0.91 (0.79, 1.04)	0.05
Cancer mortality	1.00 (reference)	0.92 (0.87, 0.97)	1.02 (0.94, 1.09)	1.10 (1.02, 1.20)	0.003
Other mortality	1.00 (reference)	0.86 (0.81, 0.93)	0.87 (0.80, 0.95)	0.94 (0.84, 1.05)	0.21

Never or occasional consumption: ≤20 g/week for men and women; low consumption: >20 g/week to ≤ 20 g/d for men, and > 20 g/week to ≤ 10 g/d for women; moderate consumption: > 20 to ≤ 40 g/d for men, and > 10 to ≤ 20 g/d for women; high consumption: > 40 g/d for men, and > 20 g/d for women

Multiple variable analyses were adjusted for: Age (continuous, years), sex (male, female), race (white, non-white), and the regions of assessment center ( Manchester, Liverpool, Newcastle, Middlesbrough, Leeds, Sheffield, Birmingham, Nottingham, Cambridge, Norwich, London, Oxford, Reading, Bristol, Plymouth, Glasgow, Edinburgh, Cardiff, Swansea), smoking status (never, former, current), education (middle school and below, high school, college or above), Townsend Deprivation Index (continuous), HEI-2020 (continuous), physical activity (continuous, MET-min/week), blood pressure (continuous, mmHg), blood glucose (continuous, mmol/l), blood cholesterol (continuous, mmol/l), family history of diabetes (yes, no), family history of CVD (yes, no), family history of cancer (yes, no), BMI (continuous, kg/m²), frailty index (continuous), alcoholic beverage preference (wine preference, other preference, no preference), alcohol drinking with meals (yes, no, varied). Alcohol consumption from wine, beer/cider, and spirits was mutually adjusted

Associations of alcohol intake from specific wine with mortality

Compared with never or occasional intake of alcohol from red wine, high intake of alcohol from red wine was associated with a 9% higher all-cause mortality and 25% higher cancer mortality; while low and moderate intake of alcohol from red wine was associated with 8–14% and 8–16% lower all-cause, CVD, and other mortality, respectively. High intake of alcohol from white wine was associated with a 16–25% higher all-cause, cancer, and other mortality; and high alcohol intake from fortified wine was associated with higher all-cause mortality (1.59 (1.24, 2.04)) and other mortality (2.41 (1.70, 3.41)), and moderate alcohol intake from fortified wine was associated with a 35% higher other mortality, compared to never or occasional intake (Supplementary Table 3).

Associations of alcohol intake with site-specific cancer mortality

Overall, total and specific alcohol intake was significantly associated with increased mortality from 15 site-specific cancer mortality outcomes (Fig. 1, Supplementary Table 4). Compared with never or occasional intake of total alcohol, high intake of total alcohol was associated with higher risk of mortality from head/neck cancers (3.01 (1.40, 6.49)); respiratory system cancers (1.79 (1.50, 2.15)), including tracheal/lung cancers (1.79 (1.50, 2.15)); gastrointestinal cancers (1.48 (1.28, 1.70)), including colorectal cancers (1.45 (1.15, 1.84)), and other gastrointestinal cancers (1.64 (1.34, 2.01))(Fig. 1-A, Supplementary Table 4).

Fig. 1.

Associations of total and specific alcohol from beer/cider, spirits and wine with various site-specific cancer mortality^&. Never or occasional consumption: ≤20 g/week for men and women; low consumption: >20 g/week to ≤ 20 g/d for men, and > 20 g/week to ≤ 10 g/d for women; moderate consumption: > 20 to ≤ 40 g/d for men, and > 10 to ≤ 20 g/d for women; high consumption: > 40 g/d for men, and > 20 g/d for women. Multiple variable analyses were adjusted for: Age (continuous, years), sex (male, female), race (white, non-white), and the regions of assessment center (Manchester, Liverpool, Newcastle, Middlesbrough, Leeds, Sheffield, Birmingham, Nottingham, Cambridge, Norwich, London, Oxford, Reading, Bristol, Plymouth, Glasgow, Edinburgh, Cardiff, Swansea), smoking status (never, past, current), education (middle school and below, high school, college or above ), Townsend Deprivation Index (continuous), HEI-2020 (continuous), physical activity (continuous, MET-min/week), blood pressure (continuous, mmHg), blood glucose (continuous, mmol/l), blood cholesterol (continuous, mmol/l), family history of diabetes (yes, no), family history of CVD (yes, no), family history of cancer (yes, no), BMI (continuous, kg/m²), frailty index (continuous), alcoholic beverage preference (wine preference, other preference, no preference), alcohol drinking with meals (yes, no, varied). Alcohol consumption from wine, beer/cider, and spirits was mutually adjusted. Cancer mortality outcomes included mortality from head and neck cancers, thyroid cancer, respiratory system cancers, laryngeal cancer, tracheal and lung cancers, gastrointestinal cancers, gastric cancer, colorectal cancer, other gastrointestinal cancers, liver and gallbladder cancers, liver cancer, gallbladder and biliary tract cancers, bone and soft tissue cancers, hematological malignancies, kidney and bladder cancers, male genital cancers, prostate cancer, female breast cancer, female genital cancers, and ovarian and adnexal cancers. ^* An FDR-corrected P < 0.05 was considered statistically significant. ⁺ indicated total cancer mortality from certain health systems. For example, gastrointestinal cancer mortality included total mortality from gastric cancers, colorectal cancers, and other gastrointestinal cancers. ^# Other gastrointestinal cancers included cancers from malignant neoplasm of esophagus, small intestine, anus and anal canal, pancreas and other digestive organs. ^& The bar that exceeded the outermost circle indicated an HR > 1.60. The specific HRs and 95%CIs are provided in Supplementary Table 4

Further, compared with never or occasional alcohol intake from beer/cider, high alcohol intake from beer/cider was significantly associated with substantially higher mortality from head/neck, respiratory system (e.g., tracheal/lung); gastrointestinal (e.g., colorectal), liver/gallbladder, hematological, and female/male genital cancers (all HRs ranging from 1.43 to 3.81). Moderate alcohol intake from beer/cider was associated with substantially higher mortality from head/neck, and respiratory system cancers (e.g., tracheal/lung) (all HRs ranging from 1.41 to 2.56) (Fig. 1-B, Supplementary Table 4). Compared with never or occasional alcohol intake from spirits, moderate and high alcohol intake from spirits was associated with substantially higher mortality from head/neck, respiratory system (e.g., laryngeal, tracheal/lung), gastrointestinal, and liver/gallbladder cancers (all HRs ranging from 1.57 to 7.99). (Fig. 1-C, Supplementary Table 4). Compared with never or occasional alcohol intake from wine, low alcohol intake from wine was significantly associated with lower mortality from respiratory system (e.g., tracheal/lung) cancers (all HRs ranging from 0.80 to 0.81); while high alcohol intake from wine was significantly associated with substantially higher mortality from respiratory system (e.g., tracheal/lung), gastrointestinal, prostate, and female breast cancers (all HRs ranging from 1.25 to 1.54) (Fig. 1-D, Supplementary Table 4).

Dose-response analysis and joint analysis results

Linear dose-response relationships of higher total alcohol intake with higher mortality from all-cause, CVD, and cancer were observed, which were mainly driven by alcohol from beer/cider and spirits in linear dose-response relationships (all P_{non−linearity}>0.05). Yet, J-shaped dose-response relationships of alcohol intake from wine were observed for all-cause and cause-specific mortality (all P_{non−linearity} <0.05) (Supplementary Fig. 2). Further, compared with drinking outside of meals, drinking with meals sometimes and usually had a 7–18% and 12–22% lower risk of all-cause and cause-specific mortality, respectively. In joint analyses, alcohol intake, together with drinking outside of meals exacerbated the detrimental associations of moderate and high beer/cider and spirits consumption and high wine use, as well as attenuated the beneficial associations of low to moderate wine use. (Supplementary Tables 5–6).

Subgroup and sensitivity analyses results

The associations of total alcohol intake with all-cause and cause-specific mortality differed by age, sex, race, smoking, physical activity, HEI-2020, or BMI (all P_interaction <0.05). For example, the associations of total alcohol intake with all-cause mortality were stronger among younger people, men, non-white people, current smoker, and non-obesity people (Supplementary Table 7). The results remained similar after excluding participants who died within the first two years of the follow-up or the lifetime abstainers (Supplementary Tables 8–9), as well as in competing risk analyses using cause-specific hazard models for CVD mortality, cancer mortality, and other-cause mortality (Supplementary Table 10).

Discussion

Primary findings

In this study, we observed that all-cause, CVD, and other mortality was significantly elevated at high total alcohol intake, and total alcohol intake was specifically associated with higher cancer mortality starting at the moderate level. The positive associations were mainly driven by alcohol use from beer/cider and spirits in a linear dose-dependent fashion, especially for even low spirits consumption in relation to higher all-cause, CVD, cancer, and other mortality. In contrast, low to moderate wine use was significantly associated with lower all-cause, CVD, cancer, and other mortality, although high wine intake was still significantly associated with higher cancer mortality. Importantly, moderate and high intake of beer/cider and spirits, and high wine use were significantly associated with higher mortality from multiple site-specific cancers, including mortality from head/neck, respiratory system, gastric, colorectal, liver, gallbladder, hematological, female breast cancer, and female/male genital cancers (e.g., prostate cancer). Drinking outside of meals tended to exacerbate the harmful associations for moderate or high alcohol use and mortality.

Compared with other studies

In contrast to most previous studies reporting protective associations of low to moderate total alcohol intake for all-cause, CVD, and cancer mortality [3, 31, 32], our study observed higher consumption of total alcohol in relation to higher all-cause, CVD, and cancer mortality in a linear fashion, particularly starting by moderate total alcohol intake for cancer mortality, and no protective associations for all-cause, CVD, and cancer mortality at any dose of total alcohol. The discrepancies between our study and prior ones could be explained by reverse causation bias [4, 5]. Most prior studies included former drinkers as the reference, despite most ceased drinking due to poor health, potentially causing reverse causation bias [33]. In our study, to minimize the bias, we excluded former drinkers and adjusted for FI, a comprehensive health index [26].

Further, existing evidence regarding the specific amount of alcohol from beer/cider, and spirits with mortality is limited. A recent meta-analysis with relatively small sample size (n < 50,000) demonstrated a U-shaped relationship between beer consumption with all-cause and CVD mortality [34]. Also, a few observational studies reported positive [15] or null associations [35] between spirits consumption with all-cause mortality. In our analysis of a much larger population, we observed that higher alcohol intake from beer/cider and spirits was linearly associated with higher all-cause and cause-specific mortality, starting by even low alcohol intake from spirits (e.g., 3 measures/week for men and women) for a 7–9% higher all-cause, CVD, cancer, and other mortality, or low alcohol intake from beer/cider (e.g., 1 pint/week for men and women) for a 7% higher cancer mortality, compared with never or occasional consumption. Only a few studies examined the associations of wine consumption with mortality, with a U-shaped or J-shaped relationship [16] or a linear inverse association [14, 36]. In our study, low and moderate wine use was significantly associated with lower mortality from CVD and other causes, whereas high wine use (e.g., ≥ 3 glasses/day for men and 1.5 glasses/day for women) was still associated with a 10% higher cancer mortality. Also, evidence regarding red, white, and fortified wine and mortality is scarce. In our study, we observed that the inverse associations for low and moderate wine use were mainly driven by red wine. Antioxidants such as resveratrol in red wine, may mainly explain lower risk of mortality for low and moderate wine use. To our knowledge, our study is the first to examine the associations of alcohol from specific wine with all-cause and cause-specific mortality and the associations of alcohol intake with 20 site-specific cancer mortality. We observed that even moderate consumption of alcohol from beer/cider and spirits, and high consumption of wine were significantly associated with substantially higher mortality from multiple site-specific cancers, especially from head/neck, respiratory system, and gastrointestinal cancers. Our results were in line with the 2025 the US Surgeon General’s Advisory, which reported alcohol drinking associated with higher risk of certain cancers (e.g., liver and breast cancers) [10], but we identified novel links between moderate and high alcohol intake with higher mortality from hematological malignancies, and female/male genital cancers, including prostate cancer, further supporting alcohol intake as a critical risk factor for cancer mortality. In addition, we observed that drinking outside of meals may exacerbate mortality caused by the amount of alcohol, irrespective of alcohol from wine, beer/cider, and spirits. Overall, our findings highlight the importance of improving alcohol consumption habits, including alcoholic beverage choices and drinking timing for health.

Strengths and limitations

The strengths of this study included the large sample size, the long follow-up, specific types of alcoholic beverages, adjustment of rich covariates, and considerations of cause-specific mortality, particularly various site-specific cancer mortality. However, several limitations should be considered. First, alcohol intake was self-reported via questionnaires and may have been subject to underreporting, leading to potential measurement error in exposure assessment. Furthermore, alcohol intake was primarily assessed at baseline and was not repeatedly measured across the full cohort during follow-up; therefore, changes in drinking behavior over time could not be fully captured. Such misclassification would likely be non-differential with respect to mortality outcomes and may have attenuated the observed associations. Second, as this UK-based study primarily included White participants, these results may not generalize to other ethnicities or populations with different lifestyles. Finally, given the observational design, we cannot fully rule out residual confounding.

Conclusions

In this study, all-cause, CVD, cancer, and other mortality were significantly elevated with high total alcohol intake. Even moderate use of total alcohol was significantly associated with higher cancer mortality. No protective associations were observed at any dose of total alcohol for all-cause, CVD, or cancer mortality. The positive associations were mainly driven by beer/cider and spirits, starting by the low to moderate alcohol intake for higher all-cause, CVD, cancer, and other mortality, particularly for cancer mortality from head/necks, respiratory system, gastrointestinal, liver/gallbladder, hematological, female breast, and male/female genital cancers. Yet, high wine intake was associated with increased cancer mortality, while low-to-moderate intake of wine was associated with lower all-cause, CVD, and other mortality. These data imply that alcohol use, particularly from beer/cider and spirits even at low to moderate levels, may not be an effective measure, but rather harmful for preventing premature deaths.

Authors’ contributions

ZC conceived the study. ZL and BT analyzed the data. ZC, TV, GZ, SZ, LW, NK, and FQ provided statistical expertise. ZC wrote the first draft of the article. All authors contributed to the interpretation of the results and revision of the manuscript for important intellectual content and approved the final version of the manuscript. ZC is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meet the criteria have been omitted.

Funding

This work was supported by Noncommunicable Chronic Diseases-National Science and Technology Major Project to Z Chen (No. 2023ZD0508604), the National Natural Science Foundation of China to Z Chen (No. 82304148), grant from the Scientific Research Program of FuRong Laboratory to Z Chen (No. 2025PT5001).

Data availability

UK Biobank data are available to all researchers for health-related research and public interest through registration at the UK Biobank (www.ukbiobank.ac.uk).

Declarations

Ethics approval and consent to participate

The National Information Governance Board for Health and Social Care and National Health Service Northwest Multicenter Research Ethics Committee approved the protocols of the UKB. Written informed consent was obtained from all participants.

Competing interests

The authors declare no competing interests.